Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-21 23:57:58 +01:00
Code Issues Packages Projects Releases Wiki Activity

Remove CTransformers support (#5807)

Browse Source
This commit is contained in:
oobabooga 2024-04-04 20:23:58 -03:00 committed by GitHub
parent 13fe38eb27
commit d423021a48
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
12 changed files with 10 additions and 163 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
README.md
View File

@ -11,7 +11,7 @@ Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.
## Features
* 3 interface modes: default (two columns), notebook, and chat.
* Multiple model backends: [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [GPTQ-for-LLaMa](https://github.com/qwopqwop200/GPTQ-for-LLaMa), [CTransformers](https://github.com/marella/ctransformers), [QuIP#](https://github.com/Cornell-RelaxML/quip-sharp).
* Multiple model backends: [Transformers](https://github.com/huggingface/transformers), [llama.cpp](https://github.com/ggerganov/llama.cpp) (through [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)), [ExLlamaV2](https://github.com/turboderp/exllamav2), [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ), [AutoAWQ](https://github.com/casper-hansen/AutoAWQ), [GPTQ-for-LLaMa](https://github.com/qwopqwop200/GPTQ-for-LLaMa), [QuIP#](https://github.com/Cornell-RelaxML/quip-sharp).
* Dropdown menu for quickly switching between different models.
* Large number of extensions (built-in and user-contributed), including Coqui TTS for realistic voice outputs, Whisper STT for voice inputs, translation, [multimodal pipelines](https://github.com/oobabooga/text-generation-webui/tree/main/extensions/multimodal), vector databases, Stable Diffusion integration, and a lot more. See [the wiki](https://github.com/oobabooga/text-generation-webui/wiki/07-%E2%80%90-Extensions) and [the extensions directory](https://github.com/oobabooga/text-generation-webui-extensions) for details.
* [Chat with custom characters](https://github.com/oobabooga/text-generation-webui/wiki/03-%E2%80%90-Parameters-Tab#character).
@ -221,7 +221,7 @@ List of command-line flags
| Flag | Description |
|--------------------------------------------|-------------|
| `--loader LOADER` | Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, AutoGPTQ, AutoAWQ, GPTQ-for-LLaMa, ctransformers, QuIP#. |
| `--loader LOADER` | Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, AutoGPTQ, AutoAWQ, GPTQ-for-LLaMa, QuIP#. |
#### Accelerate/transformers
@ -308,12 +308,6 @@ List of command-line flags
| `--checkpoint CHECKPOINT` | The path to the quantized checkpoint file. If not specified, it will be automatically detected. |
| `--monkey-patch` | Apply the monkey patch for using LoRAs with quantized models. |
#### ctransformers
| Flag | Description |
|-------------|-------------|
| `--model_type MODEL_TYPE` | Model type of pre-quantized model. Currently gpt2, gptj, gptneox, falcon, llama, mpt, starcoder (gptbigcode), dollyv2, and replit are supported. |
#### HQQ
| Flag | Description |

6
docs/04 - Model Tab.md
View File

@ -105,12 +105,6 @@ It has an additional parameter:
* **logits_all**: Needs to be checked if you want to evaluate the perplexity of the llama.cpp model using the "Training" > "Perplexity evaluation" tab. Otherwise, leave it unchecked, as it makes prompt processing slower.
### ctransformers
Loads: GGUF/GGML models.
Similar to llama.cpp but it works for certain GGUF/GGML models not originally supported by llama.cpp like Falcon, StarCoder, StarChat, and GPT-J.
### AutoAWQ
Loads: AWQ models.

1
docs/What Works.md
View File

@ -10,7 +10,6 @@
| AutoGPTQ | ✅ | ❌ | ❌ | ✅ | ✅ |
| AutoAWQ | ? | ❌ | ? | ? | ✅ |
| GPTQ-for-LLaMa | ✅\*\* | ✅\*\*\* | ✅ | ✅ | ✅ |
| ctransformers | ❌ | ❌ | ❌ | ❌ | ❌ |
| QuIP# | ? | ? | ? | ? | ✅ |
| HQQ | ? | ? | ? | ? | ✅ |

79
modules/ctransformers_model.py
View File

@ -1,79 +0,0 @@
from ctransformers import AutoConfig, AutoModelForCausalLM
from modules import shared
from modules.callbacks import Iteratorize
from modules.logging_colors import logger
class CtransformersModel:
def __init__(self):
pass
@classmethod
def from_pretrained(cls, path):
result = cls()
config = AutoConfig.from_pretrained(
str(path),
threads=shared.args.threads if shared.args.threads != 0 else -1,
gpu_layers=shared.args.n_gpu_layers,
batch_size=shared.args.n_batch,
context_length=shared.args.n_ctx,
stream=True,
mmap=not shared.args.no_mmap,
mlock=shared.args.mlock
)
result.model = AutoModelForCausalLM.from_pretrained(
str(result.model_dir(path) if result.model_type_is_auto() else path),
model_type=(None if result.model_type_is_auto() else shared.args.model_type),
config=config
)
logger.info(f'Using ctransformers model_type: {result.model.model_type} for {result.model.model_path}')
return result, result
def model_type_is_auto(self):
return shared.args.model_type is None or shared.args.model_type == "Auto" or shared.args.model_type == "None"
def model_dir(self, path):
if path.is_file():
return path.parent
return path
def encode(self, string, **kwargs):
return self.model.tokenize(string)
def decode(self, ids):
return self.model.detokenize(ids)
def generate(self, prompt, state, callback=None):
prompt = prompt if type(prompt) is str else prompt.decode()
# ctransformers uses -1 for random seed
generator = self.model(
prompt=prompt,
max_new_tokens=state['max_new_tokens'],
temperature=state['temperature'],
top_p=state['top_p'],
top_k=state['top_k'],
repetition_penalty=state['repetition_penalty'],
last_n_tokens=state['repetition_penalty_range'],
seed=int(state['seed'])
)
output = ""
for token in generator:
if callback:
callback(token)
output += token
return output
def generate_with_streaming(self, *args, **kwargs):
with Iteratorize(self.generate, args, kwargs, callback=None) as generator:
reply = ''
for token in generator:
reply += token
yield reply

29
modules/loaders.py
View File

@ -138,15 +138,6 @@ loaders_and_params = OrderedDict({
'no_use_fast',
'gptq_for_llama_info',
],
'ctransformers': [
'n_ctx',
'n_gpu_layers',
'n_batch',
'threads',
'model_type',
'no_mmap',
'mlock'
],
'QuIP#': [
'trust_remote_code',
'no_use_fast',
@ -332,13 +323,6 @@ loaders_samplers = {
'skip_special_tokens',
'auto_max_new_tokens',
},
'ctransformers': {
'temperature',
'top_p',
'top_k',
'repetition_penalty',
'repetition_penalty_range',
},
}
loaders_model_types = {
@ -348,19 +332,6 @@ loaders_model_types = {
"opt",
"gptj"
],
'ctransformers': [
"None",
"gpt2",
"gptj",
"gptneox",
"llama",
"mpt",
"dollyv2",
"replit",
"starcoder",
"gptbigcode",
"falcon"
],
}

30
modules/models.py
View File

@ -67,7 +67,6 @@ def load_model(model_name, loader=None):
'llamacpp_HF': llamacpp_HF_loader,
'ExLlamav2': ExLlamav2_loader,
'ExLlamav2_HF': ExLlamav2_HF_loader,
'ctransformers': ctransformers_loader,
'AutoAWQ': AutoAWQ_loader,
'QuIP#': QuipSharp_loader,
'HQQ': HQQ_loader,
@ -97,7 +96,7 @@ def load_model(model_name, loader=None):
shared.settings.update({k: v for k, v in metadata.items() if k in shared.settings})
if loader.lower().startswith('exllama'):
shared.settings['truncation_length'] = shared.args.max_seq_len
elif loader in ['llama.cpp', 'llamacpp_HF', 'ctransformers']:
elif loader in ['llama.cpp', 'llamacpp_HF']:
shared.settings['truncation_length'] = shared.args.n_ctx
logger.info(f"LOADER: \"{loader}\"")
@ -265,33 +264,6 @@ def llamacpp_HF_loader(model_name):
return model
def ctransformers_loader(model_name):
from modules.ctransformers_model import CtransformersModel
path = Path(f'{shared.args.model_dir}/{model_name}')
ctrans = CtransformersModel()
if ctrans.model_type_is_auto():
model_file = path
else:
if path.is_file():
model_file = path
else:
entries = Path(f'{shared.args.model_dir}/{model_name}')
gguf = list(entries.glob('*.gguf'))
bin = list(entries.glob('*.bin'))
if len(gguf) > 0:
model_file = gguf[0]
elif len(bin) > 0:
model_file = bin[0]
else:
logger.error("Could not find a model for ctransformers.")
return None, None
logger.info(f'ctransformers weights detected: \"{model_file}\"')
model, tokenizer = ctrans.from_pretrained(model_file)
return model, tokenizer
def AutoAWQ_loader(model_name):
from awq import AutoAWQForCausalLM

4
modules/models_settings.py
View File

@ -48,7 +48,7 @@ def get_model_metadata(model):
model_settings['loader'] = loader
# GGUF metadata
if model_settings['loader'] in ['llama.cpp', 'llamacpp_HF', 'ctransformers']:
if model_settings['loader'] in ['llama.cpp', 'llamacpp_HF']:
path = Path(f'{shared.args.model_dir}/{model}')
if path.is_file():
model_file = path
@ -231,7 +231,7 @@ def apply_model_settings_to_state(model, state):
loader = model_settings.pop('loader')
# If the user is using an alternative loader for the same model type, let them keep using it
if not (loader == 'ExLlamav2_HF' and state['loader'] in ['GPTQ-for-LLaMa', 'ExLlamav2', 'AutoGPTQ']) and not (loader == 'llama.cpp' and state['loader'] in ['ctransformers']):
if not (loader == 'ExLlamav2_HF' and state['loader'] in ['GPTQ-for-LLaMa', 'ExLlamav2', 'AutoGPTQ']):
state['loader'] = loader
for k in model_settings:

4
modules/shared.py
View File

@ -88,7 +88,7 @@ group.add_argument('--chat-buttons', action='store_true', help='Show buttons on
# Model loader
group = parser.add_argument_group('Model loader')
group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, AutoGPTQ, AutoAWQ, GPTQ-for-LLaMa, ctransformers, QuIP#.')
group.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: Transformers, llama.cpp, llamacpp_HF, ExLlamav2_HF, ExLlamav2, AutoGPTQ, AutoAWQ, GPTQ-for-LLaMa, QuIP#.')
# Transformers/Accelerate
group = parser.add_argument_group('Transformers/Accelerate')
@ -259,8 +259,6 @@ def fix_loader_name(name):
return 'ExLlamav2'
elif name in ['exllamav2-hf', 'exllamav2_hf', 'exllama-v2-hf', 'exllama_v2_hf', 'exllama-v2_hf', 'exllama2-hf', 'exllama2_hf', 'exllama-2-hf', 'exllama_2_hf', 'exllama-2_hf']:
return 'ExLlamav2_HF'
elif name in ['ctransformers', 'ctranforemrs', 'ctransformer']:
return 'ctransformers'
elif name in ['autoawq', 'awq', 'auto-awq']:
return 'AutoAWQ'
elif name in ['quip#', 'quip-sharp', 'quipsharp', 'quip_sharp']:

6
modules/text_generation.py
View File

@ -46,7 +46,7 @@ def _generate_reply(question, state, stopping_strings=None, is_chat=False, escap
yield ''
return
if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'CtransformersModel']:
if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model']:
generate_func = generate_reply_custom
else:
generate_func = generate_reply_HF
@ -114,7 +114,7 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
if shared.tokenizer is None:
raise ValueError('No tokenizer is loaded')
if shared.model.__class__.__name__ in ['LlamaCppModel', 'CtransformersModel', 'Exllamav2Model']:
if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model']:
input_ids = shared.tokenizer.encode(str(prompt))
if shared.model.__class__.__name__ not in ['Exllamav2Model']:
input_ids = np.array(input_ids).reshape(1, len(input_ids))
@ -128,7 +128,7 @@ def encode(prompt, add_special_tokens=True, add_bos_token=True, truncation_lengt
if truncation_length is not None:
input_ids = input_ids[:, -truncation_length:]
if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model', 'CtransformersModel'] or shared.args.cpu:
if shared.model.__class__.__name__ in ['LlamaCppModel', 'Exllamav2Model'] or shared.args.cpu:
return input_ids
elif shared.args.deepspeed:
return input_ids.to(device=local_rank)

2
modules/ui_model_menu.py
View File

@ -330,7 +330,7 @@ def update_truncation_length(current_length, state):
if 'loader' in state:
if state['loader'].lower().startswith('exllama'):
return state['max_seq_len']
elif state['loader'] in ['llama.cpp', 'llamacpp_HF', 'ctransformers']:
elif state['loader'] in ['llama.cpp', 'llamacpp_HF']:
return state['n_ctx']
return current_length

1
requirements.txt
View File

@ -68,5 +68,4 @@ https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
https://github.com/jllllll/ctransformers-cuBLAS-wheels/releases/download/AVX2/ctransformers-0.2.27+cu121-py3-none-any.whl
autoawq==0.2.3; platform_system == "Linux" or platform_system == "Windows"

1
requirements_noavx2.txt
View File

@ -68,5 +68,4 @@ https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
https://github.com/jllllll/ctransformers-cuBLAS-wheels/releases/download/AVX/ctransformers-0.2.27+cu121-py3-none-any.whl
autoawq==0.2.3; platform_system == "Linux" or platform_system == "Windows"